We found that 23% of late preterm deliveries had no recorded indication for delivery (maternal disease, obstetric complication, congenital anomaly, and spontaneous labor) on the birth certificate. The American College of Obstetricians and Gynecologists guidelines state that delivery before 39 weeks of gestation should only be undertaken when there is an accepted medical or obstetric complication or if fetal lung maturity has been documented.5
In deciding the timing of delivery for women at high risk for adverse pregnancy outcomes before term, the anticipated risk of continuing pregnancy and stillbirth should outweigh the attendant neonatal morbidity and mortality with preterm delivery.
These principles apply for late-preterm gestations with known or newly diagnosed medical, obstetric, or fetal conditions and complications. Because we found that 1 in 5 late-preterm deliveries occurred with no recorded medical or obstetric indication, we speculate that other factors may have influenced the patient and health care provider’s decision regarding the timing of delivery, namely, the perception that infants born in the late preterm period are at no greater risk for mortality and morbidity than term infants. However, confirming several recent reports,3,6–8
our data show that infants born at late-preterm gestations have significantly higher rates of mortality compared with term infants.
Whether increased rates of deliveries with no indication are the source of increasing rates of preterm births (including late-preterm births) in the United States remains unclear. To our knowledge, this is the first study to examine not only the delivery indications for late preterm births but their relationship with neonatal and infant mortality rates using recent US vital statistics data.
A novel finding in our study was that infants born at late preterm delivered without an indication had higher neonatal and infant mortality rates compared with those born after isolated spontaneous labor. It has been hypothesized that, in preterm births, labor may be “triggered” so that the fetus can exit a potentially “hostile” in-utero environment. Thus, spontaneous onset of preterm labor may be a consequence of an earlier idiopathic activation of the normal labor process or the result of a pathologic insult in an attempt to protect the fetus.9
Furthermore, the process of labor itself facilitates fetal lung maturation and improves clearance of pulmonary fluid, reducing the risk of neonatal respiratory distress.10,11
These factors may explain why the neonatal and infant mortality rates for infants born at late preterm after spontaneous labor may be lower than those for infants born at late preterm with no documented indication.
Late-preterm deliveries with an obstetric complication or major congenital anomaly had significantly higher neonatal and infant mortality rates compared with all of the other categories, including deliveries with no recorded indication, suggesting that the underlying condition that prompted the delivery was associated with poorer outcomes. Thus, it seems that survival rates for infants born at late preterm may be affected by the indication for delivery and that, within gestational age strata, all late-preterm infants are not alike with respect to their risk of death. Therefore, the decision to deliver or not during late-preterm gestations should be based on the underlying medical or obstetric factor(s) and a careful assessment of the risks of preterm delivery versus the potential benefits of expectant management.
One study that evaluated the etiology for late-preterm deliveries in a large academic center found that ~80% were attributed to idiopathic preterm labor or ruptured membranes and 20% to obstetric complications.7
Another study examined the temporal trends in preterm birth subtypes and perinatal mortality in the United States for 1989–1991 and 1995–2000.12
The authors defined medically indicated preterm births as those that followed iatrogenic intervention (labor induction or a primary or repeat cesarean delivery) and spontaneous preterm births as those that were neither associated with ruptured membranes nor were medically indicated. However, we believe that equating “primary cesarean,” “repeat cesarean,” or “labor induction” from the birth certificate files with “medically indicated deliveries” may be inaccurate. One cannot assume that a clear medical, fetal, or obstetric indication existed for preterm delivery based only on the fact that a delivery was a planned cesarean delivery or labor induction.
Some of the factors reported as contributing to the increase in late-preterm births include increasing rates of elective cesarean delivery and induction13
and increasing rates of multiple gestations.1
A recent study documented that physician practices may have changed over time. Over a 9-year period in an academic institution that had significantly lower baseline cesarean delivery rates compared with the US rates, there was a gradual increase in cesarean delivery rates paralleling their increase in late-preterm birth rates.14
These findings have significant implications as the cesarean delivery rate (24.4 in 2001 and 29.1 in 2004) and pre-term birth rate (11.9 in 2001 and 12.5 in 2004) continue to rise, indicating that the magnitude of the late-preterm birth problem will likely increase.
These factors, however, do not fully explain the dramatic increase in the rate of late-preterm births, especially for singleton gestations. Whether the increasing frequency of early cesarean deliveries is because of improved and earlier recognition of medical/obstetric indications or a greater willingness to perform a cesarean delivery for the same or even lesser indications has not been established. In fact, Declercq et al15
found no association between changes in the maternal risk profile and shifts in the primary cesarean delivery rates between 1991 and 2002, suggesting that factors other than medical or obstetric conditions may be related to increasing rates of preterm cesarean deliveries and, indirectly, late preterm births.
Our finding of an association between late-preterm deliveries with no recorded indication and social and geographic factors suggests that patient-driven factors are playing a role in this category of late-preterm births. We found that women with no recorded indication for delivery were more likely to be older, white, have higher levels of education, and live outside the Northeast. If such deliveries are occurring, it can have a major impact on the overall preterm birth rate, because non-Hispanic white women form the majority of US women of reproductive age. There are no studies assessing patient factors related to late-preterm delivery with no recorded indication, but it may be that this group of women is more likely to request that their obstetric provider deliver them before term.16,17
Consequently, patient and provider convenience factors may be contributing to the increasing rate of late-preterm delivery.
The analysis of vital statistics is of value because of the large cohort size representing the United States, comprehensiveness of births and deaths (>99%), reduced selection bias, and the ability to examine sub-populations.18
In our study, the use of linked birth-infant death files allowed us to examine the associations between maternal characteristics and subsequent infant mortality. Studies comparing birth certificate data with hospital records suggest that demographic characteristics and some medical variables, such as method of delivery, are accurately reported on the certificates, lending validity to stratification by those factors.18–23
There are limitations to using vital statistics. The inaccuracies of gestational age estimates in birth certificates are known; however, they are less frequent for late-preterm and term births.24
Underreporting for medical diagnoses, obstetric complications, and congenital anomalies is known to occur and will lead to an overestimate of the number of deliveries with no recorded indication.18–23,25–26
Although we cannot determine the magnitude of underreporting of conditions in our data set, our analyses showed that the contribution of missing data to our findings was minimal.